Energy efficiency is a good thing—but is it being undermined by some part of human nature?
There’s a long-running debate in energy economics about whether as technology becomes more efficient, people may cancel out (or significantly decrease) energy savings because they consume more resources, not fewer.
This effect, variably known as the rebound effect or the Jevons paradox, traces way back to 1865, when the English economist William Stanley Jevons noticed that as steam engines burned coal more efficiently, Britain burned dramatically more coal, not less. Cheaper energy services, he argued, simply invite more energy use.
Few examples illustrate the Jevons paradox as starkly as the humble light bulb. A modern LED produces the same brightness as a Victorian gas lamp using less than one percent of the energy (a 1,000-fold leap in efficiency). Yet humanity now uses vastly more light than ever before: glowing billboards, 24-hour parking lots, and cities visible from space. Each efficiency gain in lighting has been met, and often surpassed, by more and more lights. Did the carbon savings we expected partly evaporate into a brighter world?
The question for the climate era is uncomfortable—but unavoidable. Nearly every national climate plan, every net-zero pledge, and every IPCC pathway leans heavily on energy efficiency as a pillar of decarbonization. How much can more efficient cars, heaters, and other appliances really help stave off climate change?
• • •
The Good News
1. The big numbers look good. Energy efficiency has to date been one of the main drivers of emissions reductions. The International Energy Agency estimates that improvements to energy efficiency saved the world 7 gigatons of carbon dioxide from 2010 to 2022. For context, that’s more than the tailpipe emissions from 1.5 billion gas-powered passenger cars driven for an entire year. And, the IEA projects that improving fuel efficiency in vehicles, better insulation in houses, and other energy efficiency measures could deliver two-thirds of the oil demand reduction and half of the natural gas demand reduction necessary to meet net zero energy sector emissions by 2050.
2. Jevons’ paradox is only a problem if the metric is a problem. As Adam Dorr pointed out in a blog post for the nonprofit research org RethinkX, swapping out an emissions-heavy coal plant for a more efficient solar farm may cause energy consumption to spike as prices drop, but that doesn’t mean emissions went up. We often associate energy efficiency with energy austerity. But what if a fully decarbonized economy turned that association on its head? We could use a whole lot more energy, but our emissions footprint would be undetectable. Check out Adrienne Bernhard’s piece for the BBC on “How limitless green energy would change the world.”
Source: The International Energy Agency 2026 Electricity Report
3. A (possible) ceiling on consumption. Full-fledged rebound requires appetites without limits; in practice, energy appetites saturate. In other words, there may be a ceiling to how much energy most people actually want. A family that switches to a heat pump does not crank the thermostat to 85°F because heating got cheaper; they nudge it up a degree or two and pocket the rest. Even at a national level, at some point, enough really is—well—enough. In his New York Times article, “The Paradox Holding Back the Clean Energy Revolution,” Ed Conway cites research showing that steel and copper consumption seem to slow down as countries achieve a high standard of living.
In short, rebound may be real, but it may also be overblown within the context of carbon emissions. In fact, the strongest version of the Jevons’s claim—that efficiency raises total emissions—is, when tested against modern data, surprisingly hard to find.
• • •
The Bad News
1. AI is the wrench in the works. If ever there were a real-time Jevons experiment, it is unfolding now in data server farms in Virginia, Ireland, and Arizona. Google, for example, seems keen on energy efficiency. In their 2024 environmental report, the company reported that their latest custom processors were 2.7 times more energy efficient than the previous generation, and that they’d found ways to slash the energy required to train models by up to a thousand-fold. In their 2025 report, they highlight how improvements in hardware energy efficiency, among other things, helped them avoid two-thirds of possible emissions the previous year. And yet, that same report noted that once you include the emissions produced building and rigging up their new AI data centers, Google’s overall real-world emissions have actually risen by more than 50% between 2019 and 2024. AI systems overall were estimated to have had the same carbon footprint as New York City in 2025.
Source: The International Energy Agency 2026 Electricity Report
2. Shipping may also have a big rebound. A 2024 study in Nature Energy found that Jevons’ may have eroded the carbon savings from regulations designed to increase fuel efficiency in long-haul trucking by more than 25 percent. “We didn’t anticipate effects of this magnitude,” Jonathan Hughes, one of the study’s authors, told Anthropocene. That’s because more fuel efficient trucking is cheaper trucking, which could encourage manufacturers to switch from the relatively cleaner, but slower rail shipping.
3. Rebounds don’t stay in one lane. If increases in energy efficiency result in less demand from power plants for petroleum to burn, one might think this would result in a straightforward reduction in petroleum use, but not so fast. As investment management firm Van Eck pointed out in a blog post, petroleum isn’t just an energy source, it’s also a feedstock for many petrochemicals such as plastics and fertilizers. If increased energy efficiency drives down petroleum demand, basic economics suggests petroleum prices should also go down. Manufacturers might happily gobble up the cheaper feedstocks to produce more plastics and fertilizers. Considering that petrochemicals also produce emissions (around 5 percent of the US’ annual emissions), what had been a simple picture gets messier.
• • •
What to Keep An Eye On
1. Autonomous vehicles. When researchers conducted a full life-cycle analysis of autonomous electric cars, they found some tell-tale signs of rebound. While autonomy cuts fuel-use emissions by about 21%, manufacturing the more complex vehicles, combined with increased travel, can surge emissions by up to 40%—and even with recycling offsetting some of that, autonomous electric vehicles end up emitting roughly 8% more greenhouse gases over their lifetime than standard electric vehicles.
2. The Global South. Most rebound studies come from wealthy economies where appetites for light, heat, and mobility are largely saturated. In countries where billions of people are only now gaining reliable electricity, air conditioning, and personal vehicles, even modest efficiency gains may unlock enormous new demand. How the world handles that legitimate growth, and whether the energy meeting it is clean, may matter more for the global carbon trajectory than any rebound coefficient ever measured.
3. Carbon pricing. Even where rebound is real, it is not destiny—it is a policy problem with a known fix. Inês Azevedo’s makes the point in a 2014 paper. When efficiency is paired with a carbon price, an emissions cap, or a clean-electricity standard, the freed-up money and energy cannot simply re-fuel fossil consumption, because the cap or the price is still binding. Efficiency under a carbon constraint is not Jevons’s coal mine; it is a tightening lid on a shrinking budget. The paradox, in this view, is not a law of human nature—it is what happens when you do efficiency without doing climate policy.
